The cavities in this Decibel DB4076Z duplexer are made of unplated aluminum. Decibel also sells another version, the DB4076W, which uses copper construction. The copper DB4076W has less insertion loss and better isolation than the aluminum DB4076Z. This is mainly caused by the higher resistivity of aluminum and its surface oxidation. The increased losses degrade the Q of the cavity resulting in a response that is less "sharp". I don't particularly like unplated aluminum cavities for this reason, so I didn't mind cannibalizing an otherwise good duplexer to get a clean-up cavity for this link transmitter. Current-vintage DB4076Z duplexers ARE copper; the Z suffix now means that they are made for wide-split use whereas previously the Z indicated that they were aluminum. A little confusing I know.
There are quite a few variations of the DB4075/DB4076 series duplexers. The DB4075 is the same as a DB4076, except only three cavities (2 on Tx, 1 on Rx) as compared to the 4-cavity DB4076. The DB4075/6 duplexers were OEM'ed for several companies, including EF Johnson for use in the CR1000 line of repeaters, as well as for General Electric for the MASTR II series.
The original version of the cavities used in the DB4075/6 were squarish and light tan in color. Subsequent versions were square and a darker tan color. Both versions of the square cavities used two loops per cavity with a shunt reactance (variable capacitor) between the loops' hot ends. Both versions used copper cavity construction.
There were several types of loop assemblies for these square-vintage cavities which made the cavities perform differently. Some were asymmetrical in that they are optmized for either high-pass/low-reject use or vice-versa. Others can be used for either high-pass or low-pass with similar results. Typically, if an asymmetrical cavity is used for the "wrong" side of a duplexer, it may appear to work properly in that you are able to set the pass and reject frequencies where they should be, and it may even yield adequate isolation, however, return loss will not be optimized to match pass response. In other words, tuning the cavity will always be a compromise between a good match and least loss. If you have a duplexer of this type, you might find it an educational exercise to sweep the cavities for both pass response as well as return loss. Also try sweeping in both directions by swapping the cables going to the loop connectors. You will likely find that they a) don't perform the same in high-pass versus low-pass configurations, and b) that which loop is used as the "input" versus "output" also alters the behavior, sometimes slightly, sometimes greatly. You may find that one direction tunes better than the other in terms of matching return loss and insertion loss optimizations.
Although this may seem strange at first, there is a perfectly good reason why this is the case. The mechanical construction of the capacitors used in some of the duplexers requires that the connection to the two loops not be identical. The lead connecting the bottom of the capacitor to one loop is longer in length and further away from the wall of the cavity, making it more inductive than the other lead which has a shorter path and is closer to the cavity wall. To make a long story short, the the construction of the loop assemblies isn't mechanically symmetrical, and likewise not electrically symmetrical, which affects the source and load impedance seen from the outside world at the loop connectors, as well as the coupling and resonance/anti-resonance of the cavity filter as a whole. But I digress..
Later versions of the DB4075/6 use a round cavity painted tan, then later a medium-dark grey. There were two different electrical designs of these cavities. One version used two loops per cavities with a shunt reactance across the loops. Another version used a single loop with a series reactance. This latter type of cavity requires a tee connector to be installed on the loop connector. This is the type of cavity (single loop) being worked on in this article. Like the earlier square cavities, the cable harness varied depending on the type of loop(s) used.
Finally, the modern version of the DB4075/6 duplexers use a round cavity, two loops with a shunt reactance, and are painted Decibel "Tek Black".
As you might expect, there are many versions of the cable harnesses for all of these duplexers. Very early versions used a double-shielded cable similar to RG142/B but with NO jacket. The silver-plated braid was left exposed to the air. Normally this exposed braid wouldn't be a big concern in an indoor environment, but because the interconnect cables were coiled into loops, and adjacent loop turns were left to rub together, and also rub against other parts of the duplexer (the tuning rods, the connectors, adjacent interconnect cables, etc.), there was a BIG problem with duplex noise with these harnesses. Later versions used jacketed RG142/B or RG400 interconnect cables as is now standard.
The interconnect cable lengths likewise varied with cavity and loop designs. Keeping track of what cable length goes with what version duplexer is a nightmare, especially considering you can't always tell what vintage cavities you're dealing with without opening them up. And, as mentioned before, the cable lengths between cavities, as well as between the first cavities and the tee connector, also varied in relation to the asymmetrical behavior of the cavities. That is, the cable lengths on the high-pass half of the duplexer aren't the same as the low-pass half. If you have a Decibel catalog handy, take a look at picture of a current-vintage DB4076 and you'll see what I mean.
OK, this wasn't supposed to turn into a Decibel cavity trivia dissertation, so I'm going back to finish the main guts of the article.